The company that built Southeast Asia's super-app is now building with AI agents, and they're sharing the playbook.
The Summary
- Sea Limited's CPO David Chen details how the company deployed OpenAI's Codex across engineering teams to accelerate what he calls "AI-native software development" in Asian markets
- Sea — which runs Shopee, Garena, and SeaMoney across 600M+ users — is betting that agentic coding tools create competitive advantage in speed-sensitive markets
- The shift isn't just about faster code: it's about reorganizing engineering teams around human-agent collaboration patterns that didn't exist 18 months ago
The Signal
Sea Limited doesn't do pilot projects. When David Chen says they're deploying Codex across engineering teams, he means the infrastructure layer that serves Southeast Asia's largest e-commerce platform, its gaming division, and a digital bank. The company is using AI coding agents to compress development cycles in markets where being two weeks late can mean losing a country.
Chen's framing matters here: "AI-native software development." Not AI-assisted. Not AI-enhanced. Native. He's describing a fundamental reorganization of how software gets built when agents are first-class team members. For Sea, this means agents writing boilerplate, debugging edge cases, and generating test suites while human engineers focus on architecture decisions and product intuition that agents can't yet replicate.
"Speed-sensitive markets demand AI-native development patterns that treat agents as infrastructure, not assistants."
The Asian market context amplifies everything. Sea operates across Indonesia, Thailand, Vietnam, the Philippines, Malaysia, Singapore, and Taiwan. Different languages, payment systems, regulatory frameworks, consumer behaviors. The complexity that used to require massive engineering teams now gets compressed through agent-augmented development. Chen isn't claiming agents replace engineers — he's saying they change what engineering teams optimize for.
What Sea is actually testing: can agents help a platform company move fast enough to out-execute both local competitors and Western tech giants trying to enter Asian markets? The answer shapes whether agentic coding tools become table stakes for every growth-stage tech company, or stay confined to the companies that can afford OpenAI's enterprise tier.
The Implication
Watch how Sea's competitors respond in the next two quarters. If agentic development creates meaningful speed advantages in high-velocity markets, every platform company in Asia will need a Codex-equivalent deployment plan. If it doesn't, we'll learn that agents still can't handle the cultural and regulatory complexity that makes Asian market expansion so expensive.
For engineers: the skill that matters isn't writing more code faster. It's knowing which problems to hand to agents and which ones require human judgment about markets, users, and systems. Sea is figuring out that boundary in real time, at scale, in markets where failure is public and fast.